Photolithography using a patterned masking layer is commonly used for formation of electronic components such as semiconductor integrated circuits. A pattern is transferred to a reactive photoresist layer on the semiconductor substrate by exposure of the patterned masking layer overlying the substrate to light or other energy source (such as ultraviolet light or electron beam radiation) which passes through open areas of the patterned masking layer.
The masking layer can be, for example a photomask or reticle, prepared from a photomask blank that typically comprises a thin layer of an opaque or nontransparent material, typically a metal-containing material such as a chromium-, molybdenum- or tungsten-containing material, deposited on a transparent plate of glass or quartz by sputtering or vacuum evaporation, for example. A pattern can be formed on the photomask blank structure by standard photolithography processes to selectively remove portions of the nontransparent layer (e.g., opaque chromium film) and create open areas. The pattern can be etched, for example, by forming a photoresist material over the nontransparent layer, forming a pattern in the photoresist material with an electron beam or laser beam, and transferring the pattern to the underlying nontransparent material with an etch that removes exposed portions of the nontransparent material.
With the fast pace of reduction of the micro-lithography design rule and the significant increase of data volume, the requirement on critical dimension (“CD”) control and throughput on a photomask manufacture is becoming more and more challenging. Currently, photomasks are patterned with either 50 kV e-beam or laser pattern generation tools, and chemically amplified photoresists (“CA resists”) that are sensitive to shorter wavelengths of light are being used to reduce the beam exposure (dose) and settling times, thereby improving the throughput of photomask pattern exposure. CA resists are more sensitive to radiation than non-chemically amplified resists such as ZEP-7000, thus needing a smaller dose of radiation (e.g., 8-9 μcoulomb versus about 25 μcoulomb) and a shorter development time.
Typical CA resists are composed of an acid generator that produces acid upon exposure to radiation (e.g., ultraviolet (UV) light, laser light, X-rays, electron beam), and acid-labile compounds or polymers that have changed solubility in a developer solvent through acid-catalyzed reactions. The irradiated area forms an image in the CA resist, and an annealing step (post exposure bake, or PEB) is performed after exposure. The photoresist is then exposed to a developing solvent to remove portions of the resist, the underlying opaque chrome or other nontransparent layer is etched, and the CA resist is removed. The patterned photomask is used to transfer a pattern to a photoresist layer disposed on a wafer during a fabrication step. The patterned photomask layer is exposed to blanket radiation that passes through open areas of the photomask layer onto the surface of the photoresist. The photoresist is then developed providing open areas exposing a material layer on the wafer. A fabrication step, for example, an ion implantation or an etch of the material layer, can be performed to form an underlying semiconductor device structure.
Ideally, it is advantageous to coat the photomask blank with a CA resist and immediately expose the resist on the blank. However, in practice, resist-coated blanks are often stored for several hours up to weeks before exposure. A difficulty with the use of CA resists is the change in stability and radiation dose sensitivity that occurs when a CA resist is coated on a substrate that is left standing inside or outside an irradiation apparatus. This effect is commonly known as “post coat delay” (PCD) effect.
Development of CA resists depends on the catalytic action of the acid generator, and if acid loss (neutralization) occurs, which decreases the radiation dose sensitivity of the resist, abnormalities will occur during pattern formation resulting in a significant difference between critical dimensions (CD) (i.e., minimum feature dimensions) of the patterned features compared to the required pattern dimensions. The acid loss is believed to be due to the presence of nitrogen-containing compounds such as atmospheric ammonia or ammonium ions, which neutralize the acid generator resulting in incomplete solubility of the resist, thus limiting shelf life of a photomask blank with an applied CA resist layer.
Typically, to compensate for the loss in sensitivity, the amount of decrease in dose sensitivity over time is estimated and the exposure process is adjusted accordingly. The relative dose sensitivity of the CA resist layer can be determined through experimentation and/or theoretical calculation. The difference in dose sensitivity may be a linear function of time difference between exposure of regions, a logarithmic difference, or related by some other mathematical correlation, depending on the type of radiation utilized, the type of resist utilized, and the tolerance for differences in minimum feature size between a first blank and a second blank.
Several methods have been used in an attempt to extend the shelf time or length of time a photomask substrate with a CA resist applied can be stored prior to exposure to the imaging/patterning radiation. Such methods have included the use of protective coatings and treatment with oxygen plasma to reduce the exposure of a CA resist to nitrogen-containing compounds. While use of these known methods produces some improvement over an untreated substrate, significant profile defects are still evident.
It would be desirable to develop a process for reducing the PCD effect on a chemically amplified (CA) resist coating disposed on a substrate such as a photomask blank.